Skip to main content

​
SLM Evaluator

The admin approval flow launches a Docker container to run the SLM evaluator pipeline.

​
Local Dev Setup

Build the Docker image once before using it: 
docker build -t slm-evaluator:latest ./pipeline/
Rebuild this image after any changes to pipeline/slm_evaluator.py or pipeline/requirements.txt. Ensure PIPELINE_LOCAL_MODE=true is set in your .env (it is by default in .env.example).

​
Admin Publish Flow

After evaluation completes, the admin publish flow involves two steps:
  1. Publish to Gateway — registers the model on the internal API gateway (available via neurometric/<slug> for testing).
  2. Publish to Marketplace — creates the public marketplace listing.
Both steps can be individually reversed via the “Unpublish from Gateway” and “Unpublish from Marketplace” actions in the admin dashboard.